以相关性确定条件属性的概化决策树

在介绍了一些典型决策树分类算法的基础上,研究了一种基于相关性分析的决策树分类器。其主要思想是通过属性相关性来压缩训练集的大小并在建立决策树过程中采用此度量值来确定划分条件属性的顺序,通过阈值设定和处理简化了决策树的剪枝和优化过程,提高了处理的效率和规模。文章详细描述了算法的执行过程以及正确性证明和时间复杂性分析。

A Generalized Decision Tree Using Relevance Analysis to Evaluate Condition Attributes

Efficiency and scalability are fundamental issues concerning data mining in large databases.The decision tree is an important classifier in data mining.In this paper a decision tree classifier based on relevance analysis is proposed after discussing traditional algorithms.The main idea is to compact the training data and evaluate condition attributes with correlations,and made pruning and optimization process simplified in order to get high accuracy and fast classifying speed,which leads to efficient,high-quality,multiple-level classification of large amounts of data.The accuracy of the algorithm was proven and the complexity of time was analyzed in the paper,too.